The Euler-Maupertuis principle of least action as
variational inequality
Remco Leine, Ueli Aeberhard
To cite this version:
Remco Leine, Ueli Aeberhard. The Euler-Maupertuis principle of least action as variational
inequality. Sixth International Congress on Industrial Applied Mathematics (ICIAM07) , Jul
2007, Zürich, Switzerland. 7 (1), Proceedings in Applied Mathematics and Mechanics. .
HAL Id: hal-01310672
https://hal.archives-ouvertes.fr/hal-01310672
Submitted on 3 May 2016
HAL is a multi-disciplinary open access
archive for the deposit and dissemination of scientific research documents, whether they are published or not. The documents may come from
teaching and research institutions in France or
abroad, or from public or private research centers.
L’archive ouverte pluridisciplinaire HAL, est
destinée au dépôt et à la diffusion de documents
scientifiques de niveau recherche, publiés ou non,
émanant des établissements d’enseignement et de
recherche français ou étrangers, des laboratoires
publics ou privés.
Distributed under a Creative Commons Attribution - ShareAlike 4.0 International License
The Euler-Maupertuis principle of least action as variational inequality
R. I. Leine∗1 and U. Aeberhard1
1
Institute of Mechanical Systems, Department of Mechanical and Process Engineering, ETH Zurich, CH-8092 Zurich,
Switzerland.
Starting from Hamilton’s Principle, the current paper discusses how we can derive the Euler-Maupertuis Principle of Least
Action in the context of non-smooth dynamics. This variational principle allows us to directly obtain the space curve y(x)
of a point-mass in a potential field V (x, y) without referring to the temporal dynamics. This paper generalises the EulerMaupertuis Principle of Least Action to systems with impact by formulating the principle as a variational inequality.
1 Introduction
In 1744 Euler stated that the motion of a point-mass m in potential field V is described by the variational principle
x1 √
2m
E − V (x, y) 1 + y (x)2 dx = stationary,
E = T + V = const.,
x0
where x and y are the coordinates of the point mass in an inertial frame. This principle is called the (Euler-Maupertuis)
Principle of Least Action and is in this form sometimes known as the Principle of Jacobi. The aim of this paper is to put the
Principle of Least Action in the context of non-smooth dynamics, giving the principle a novel application field.
2 Principle of d’Alembert-Lagrange as variational inequality
Let us consider a point-mass m which we can push with an external force F against a frictionless wall, or even into the corner
of a wall. The wall reacts with a contact force λ. Newton’s second law
mr̈ = F + λ
(1)
gives the equilibrium between the inertia force, the external force F and the contact force λ. The contact force λ for a
frictionless wall is an element of the normal cone,
−λ ∈ NK (r),
(2)
where the normal cone NK (r) can be expressed as the subderivative ∂ΨK (r) of the indicator function on the domain K
which is enclosed by the wall. For simplicity we will assume K to be convex. If we substitute the equation of motion (1) in
the set-valued force law (2), then we obtain a differential inclusion
−(mr̈ − F ) ∈ ∂ΨK (r)
(3)
for the non-impulsive dynamics of the point-mass with unilateral constraint. Using some concepts of convex analysis we
obtain the following result:
ΨK (r ∗ ) ≥ ΨK (r) − λT (r ∗ − r) ∀ r ∗
0 ≥ −λT (r ∗ − r) ∀ r ∗ ∈ K,
T
T
λ δr = (mr̈ − F ) δr ≥ 0
δr = r ∗ − r
(4)
∀ δr ∈ TK (r)
The virtual work λT δr of the contact force λ is nonnegative for all virtual displacements which are taken from the tangent
cone TK (r), which is the cone of all kinematically admissible virtual displacements. The expression (mr̈ − F )T δr ≥ 0,
∀ δr ∈ TK (r) is called a variational inequality.
Consider the principle of d’Alembert-Lagrange for a perfect bilateral constraint: the virtual work of a perfect bilateral
constraint force vanishes for all kinematically admissible virtual displacements. The variational inequality which we just
obtained is nothing else than the principle of d’Alembert-Lagrange for a perfect unilateral constraint.
1
3 Principle of Hamilton in inequality form
We now integrate the principle of d’Alembert-Lagrange as variational inequality over a compact time-interval I = [t0 , t1 ].
This time-interval can contain collision times which we gather in the set D = {ti }. For the moment, we leave the impact
times aside and we integrate over the interval I\D:
(mr̈ − F )T δr dt ≥ 0
∀ δr ∈ TK (r).
(5)
I\D
As is customary in this kind of analysis we perform a partial integration.
t1 t+
T
i
mṙδr t0 +
mṙδr t− −
mṙ δ ṙ dt −
F T δr dt ≥ 0
i
i
I\D
I\D
The boundary terms for t0 and t1 vanish for fixed boundary conditions r(t0 ) = r0 , r(t1 ) = r1 . The boundary terms at
the collision times D vanish if the so-called Weierstrass-Erdmann conditions are fulfilled. The first Weierstrass-Erdmann
condition requires that the impulsive forces lie within the normal cone, −Λ ∈ NK (r). The second Weierstrass-Erdmann
condition demands that the collisions are completely elastic such that there is no energy loss during the impact. The second
condition is superfluous if we consider a special class of test functions which lead to Gâteaux derivatives (see [1]). The
remaining terms can be recognised as the variation of the kinetic energy δT = mṙ T δ ṙ and the potential energy δV = F T δr,
i.e.
−
δT dt +
δV dt ≥ 0.
(6)
I\D
I\D
We are still integrating over the time-interval without collision times. However, the missing integration parts are Lebesgue
integrals and the collision times are Lebesgue negligible:
δT dt = 0
δV dt = 0.
(7)
D
D
We obtain the principle of Hamilton as variational inequality,
−δ L dt ≥ 0
L := T − V ∀ δr ∈ TK (r),
(8)
I
which also accounts for perfect unilateral constraints (see [1]).
4 Principle of Least Action in inequality form
In the same way we are also able to generalise the principle of least action to unilateral constraints. We can do so by starting
from the principle of Hamilton (8) in inequality form and by taking the conservation of energy E = T + V as auxiliary
condition. This relates the variation of the potential energy δV = −δT to the variation of the kinetic energy. Substitution of
the latter in the principle of Hamilton gives a Lagrange function which is 2 times the kinetic energy
−2 δT dt ≥ 0 ∀ δq ∈ TK (q),
in which we used generalised coordinates q. A reformulation of the kinetic energy
2
1
dθ
1
T = q̇ T M (q)q̇ = q T M (q)q 2
2
dt
gives the principle of least action in inequality form:
√ − 2δ
E0 − V (q) q T M (q)q dθ ≥ 0 ∀ δq ∈ TK (q).
Once again, we obtain a variational inequality.
References
[1] R. I. Leine, U. Aeberhard, and Ch. Glocker. ”Hamilton’s principle as variational inequality for mechanical systems with impact”, to be
submitted, 2008.
2